A substantial fraction of the automotive assembly comprises formed sheet metal parts. To reduce vehicle weight and improve fuel economy, total sheet metal mass should be minimized without compromising the structural integrity of the vehicle. Excessive deformation contributes to tearing or buckling of the metal, and therefore a forming limit is investigated experimentally to determine the extent to which each particular material can be safely strained. To assess sheet metal formability, this thesis proposes a novel framework for sheet metal surface strain measurement using a scalable dot-grid pattern. Aluminum sheet metal samples are marked with a regular grid of dot-features and imaged with a close-range monocular vision system. After forming, the sheet metal samples are imaged once again to examine the deformation of the surface pattern, and thereby resolve the material strain. Grid-features are localized with sub-pixel accuracy, and then topologically mapped using a novel algorithm for deformation-invariant grid registration. Experimental results collected from a laboratory setup demonstrate consistent robustness under practical imaging conditions. Accuracy, repeatability, and timing statistics are reported for several state-of-the-art feature detectors. / Thesis / Master of Applied Science (MASc)

<p>Accurate measurement of surface grids through imaging enables a variety of applications. One important example can be found in automotive manufacturing, where deformed sheet metal surface strains must be validated in safety critical regions, and rapidly measured to correct process variations. This thesis advances machine vision techniques in the context of close-range surface imaging and measurement. Sheet metal surface strain analysis provides a motivating application, but the contributions may be directly transferred to a variety of other machine vision applications where reliable, accurate measurements are required in adverse imaging conditions.</p> <p>Close-range imaging in practical environments presents a number of challenges, primarily relating to depth of field blur and the regional field of view. This thesis contributes to three major components required for close-range optically-based surface measurement. First, an approach for grid line intersection measurement in the presence of significant and varying depth-of-field blur is considered, with a solution based on scale-space ridge extraction. An architecture for acceleration of the computationally intensive algorithm is then developed, and implemented using state of the art graphics (GPU) hardware. Acceleration to camera video frame rates is achieved.</p> <p>The second contribution is a novel approach for interframe motion tracking of uniform gridded surfaces. The algorithm exploits topological structure of the imaged grid pattern, thereby reducing dimensionality of the interframe tracking problem. Intrinsic fiducial measurement is proposed to avoid the need for explicit feature detectors that locate fiducials in the presence of varying size and blur. Close-range interframe tracking is demonstrated, and statistics are presented on the registration objective function.</p> <p>Finally, an approach is considered for camera and hand-eye calibration of a monocular camera mounted to the tool point of a coordinate measuring machine (CMM). Pre-processing algorithms are contributed to prepare close-range gridded image data for the calibration process. Ideal model coordinate points are coherently assigned to detected grid features across video sequences, and grid approximation is performed for highly blurred image frames where reliable features have not been extracted.</p> <p>The contributions of this thesis make significant progress toward enabling video frame rate, close-range, computer vision-based sheet metal surface strain analysis, and other applications where challenging image conditions impede measurement.</p> / Doctor of Philosophy (PhD)